Trained on 16 highly edited images with expert level tagging. Epoch 12 { "lr": 0.0001, "steps": 1536, "engine": "ai-toolkit", "epochs": 12, "batchSize": 1, "ecosystem": "anima", "keepTokens": 0, "networkDim": 32, "resolution": 1024, "lrScheduler": "cosine", "minSnrGamma": null, "noiseOffset": 0.03, "networkAlpha": 16, "optimizerType": "adamw8bit", "shuffleTokens": false, "textEncoderLr": null, "flipAugmentation": true, "trainTextEncoder": false }
Description
I make sure updating this lora is unnecessary and have thoroughly tested it with multiple seeds.
Comments (3)
Awesome style!
Thanks. I'll try to make 3 new style Lora's a week. I'm currently finishing Xxoom, Tixnen, and Midnight Maid. I have data for Milo Manara, Azumaarts, Blimaxxx, Zpark, AoranPB, Arttoru, Ayaki, B O Rukia, Beuto, Tite Kubo, Cirenk, Cryosurce, dairoku tenma, equinox006, ickpot, popogori, Virgo art, and vongrauenhausen that I still need to edit prior to tagging.
@ARTREFINER Good luck.